CN103018181A - Soft measurement method based on correlation analysis and ELM neural network - Google Patents
Soft measurement method based on correlation analysis and ELM neural network Download PDFInfo
- Publication number
- CN103018181A CN103018181A CN2012105416673A CN201210541667A CN103018181A CN 103018181 A CN103018181 A CN 103018181A CN 2012105416673 A CN2012105416673 A CN 2012105416673A CN 201210541667 A CN201210541667 A CN 201210541667A CN 103018181 A CN103018181 A CN 103018181A
- Authority
- CN
- China
- Prior art keywords
- correlation analysis
- data
- neural network
- near infrared
- analysis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The invention discloses a soft measurement method based on correlation analysis and an ELM neural network. The method comprises the steps of: collecting near infrared spectrum data of a product in a solid fermentation process, analyzing and converting a collected spectral signal by a spectrograph and then transmitting the spectral signal into a computer through a data wire; carrying out pretreatment on the obtained original spectrum data; repeating the experiment for N times, selecting a bath of fermentation processes to monitor the near infrared spectrum data as case sample data; carrying out correlation analysis by a statistical correlation analysis method and other (N-1) bathes of data, and associating with actually measured reference value of product parameter indexes in the solid fermentation process by correlation index analysis results; and building a soft measurement model based on ELM. The soft measurement method is simple and convenient to operate, rapid in detection speed and good in reproducibility, and can be used for online monitoring of the quality of the product in the solid fermentation process. The soft measurement method is expected to solve the problems of high cost, long consumed time, low efficiency and the like of a common offline physical and chemical detection method in the solid fermentation production process.
Description
Technical field
The present invention relates to a kind of flexible measurement method based on correlation analysis and ELM neural network, belong to the solid ferment process control field.
Background technology
Solid state fermentation (solid-state fermentation, SSF) refers to the process of cultivation microorganism in not containing or contain hardly the wet solid material of free water.Solid-state fermentation process parameter is take pH as example: pH is an important factor in the sweat, each microorganism has one to be fit to its growth and the active pH scope of performance, control or a problem that waits to solve of pH in the present solid state fermentation, heterogeneity in the sweat constantly changes pH on the one hand, is owing to the pH that does not have in the suitable definite solid-state material of together detection on the other hand.PH in many solid ferment process has distinctive variation, says that just water cut lower in the material makes pH detection method routinely be difficult to prove effective, thereby has limited the feasibility of pH as important control parameter.In addition, also all with these two important procedure parameters closely contact is arranged as parameters such as biomass concentration and purpose product content.
At present, all adopt off-line chemical experiment method the detection of solid-state fermentation process parameter (such as humidity, pH, biomass concentration).Although the result of chemical detection method is objective credible, because the shortcoming such as its step is loaded down with trivial details, detection time is long, testing cost is high, and off-line measurement has brought a lot of inconvenience to control and the optimization of Fermentation Engineering.Therefore, be unfavorable for realizing optimal control to whole sweat status information variable.Near infrared spectrum (Near Infrared Spectroscopy, NIR) analytical technology has fast, can't harm, accurately, the advantages such as polycomponent detects simultaneously, be one of mature technology that is suitable for implementation in most line analysis and in real time control, be widely applied in fields such as oil, chemical industry, food, pharmacy and tobaccos.
Summary of the invention
The objective of the invention is: for solid-state fermentation process parameter detection method above shortcomings in the prior art, on the basis of near infrared spectrum data, provide a kind of solid-state fermentation process parameter soft measurement method based on correlation analysis and ELM neural network.
Technical scheme of the present invention is:
Flexible measurement method based on correlation analysis and ELM neural network, take different batches solid ferment process sample near infrared spectrum data correlation factors as the soft-sensing model input variable, take the actual measurement reference value of solid ferment process product parameter index as output variable, adopt the ELM neural network to carrying out solid ferment process key parameter soft sensor modeling; The step of described method is:
1) utilize diffuse reflection type near infrared spectra collection device to obtain the near infrared spectrum data of solid ferment process product, the spectral signal that is gathered imports computing machine into by data line after the spectrometer analysis conversion;
2) the original spectrum data that obtain are carried out pre-service, above-mentioned experiment is repeated N time, namely obtain the N batch data, choosing a collection of sweat monitoring near infrared spectrum data is the case sample data;
3) then adopt statistics correlation analysis and other (N-1) batch data to carry out correlation analysis;
4) utilize correlativity index analysis result and the actual measurement reference value of solid ferment process product parameter index to carry out related, set up the soft-sensing model based on ELM.
Further, described ir data correlation factors analytical approach is chaos time sequence mutual correlation dimensional analysis method.
Further, the concrete gatherer process of described step 1) is: collect N batch fermentation batch, different fermentations solid ferment process product sample constantly is used for carrying out model tuning, each sample takes by weighing puts into sample cup (spectrometer standard fitting) about 40g, and places it on the objective table; Near infrared spectrometer is connected with objective table by y-type optical fiber, and the spectral signal of collection imports near infrared spectrometer into by y-type optical fiber, is reached in the computing machine by the data line that is connected between computing machine and the spectrometer again.
Further, described step 2) preprocess method in comprises standard normal variable conversion, level and smooth, centralization, differentiate, normalization and Wavelet Denoising Method, described preprocess method can be the independent utilization of a certain method in the described preprocess method, also can be that the combination of several method is used.
Further, the actual measurement reference value in the described step 4) is measured by the conventional physical and chemical analysis method.
Further, with reference to the concerned countries standard, record the reference measurement values of solid ferment process product parameter index by the physico-chemical analysis method, form a database, described parameter index comprises biomass content and/or protein content and/or humidity and/or PH.
The invention has the beneficial effects as follows:
The present invention compares with the traditional chemical analysis means, fast and the favorable reproducibility of simple to operation, detection speed, can be used for the on-line monitoring of solid ferment process product quality, as a kind of quality control method that has application prospect, the present invention is expected to solve in the solid state fermentation production run problems such as conventional high, the consuming time length of off-line physics and chemistry detection method cost and efficient is low.
Description of drawings
Fig. 1 is technical scheme schematic diagram of the present invention;
Fig. 2 is the structural representation of operative installations of the present invention.
Among the figure: 1, sample cup; 2, objective table; 3, y-type optical fiber; 4, computing machine; 5, data line; 6, near infrared spectrometer.
Embodiment
The present invention is on the basis that solid ferment process sample analysis near infrared spectrum data is analyzed, a kind of soft-measuring modeling method based on correlation analysis and ELM neural network is provided, can satisfy simultaneously the needs of the real-time detection of multi-target ingredient, help to realize solid ferment process is carried out Real Time Monitoring and diagnosis, can guarantee the quality of final fermented product.
At first, utilize diffuse reflection type near infrared spectra collection device to obtain the near infrared spectrum data of solid ferment process product, the spectral signal that is gathered imports computing machine into by data line after the spectrometer analysis conversion; Then, the original spectrum data that obtain are carried out pre-service.Above-mentioned experiment is repeated N time, namely obtain the N batch data.Choosing a collection of sweat monitoring near infrared spectrum data is the case sample data, then adopt statistics correlation analysis and other (N-1) batch data to carry out correlation analysis, it is related that recycling correlativity index analysis result and the actual measurement reference value (being measured by the conventional physical and chemical analysis method) of solid ferment process product parameter index are carried out, and foundation is based on the soft-sensing model of ELM.
Solid ferment process product key parameter flexible measurement method based on correlation analysis and ELM neural network is by gathering the near infrared spectrum data of sweat product sample, setting up the soft-sensing model of solid ferment process key parameter index in conjunction with physico-chemical analysis methods and results, correlation analysis and ELM neural net method again.Sample to be tested by corresponding spectrum data gathering, the pre-service of original spectrum data and with the correlation analysis of case sample data, the soft-sensing model that recycling has established is predicted the property value of this sample key parameter index.
The present invention has versatility to the fast detecting of solid ferment process product parameter index, can be as follows with reference to the method for this embodiment:
Example performing step of the present invention is consulted Fig. 1, and the example implement device is consulted Fig. 2.The implementation step is as follows:
In order to eliminate inconsistent etc. the impact of background interference, grain size and uniformity coefficient, improve the quality of spectrum, need the original spectrum data that gather are carried out pre-service, the preprocess method of spectrum mainly contains standard normal variable conversion, level and smooth, centralization, differentiate, normalization and Wavelet Denoising Method etc., in these preprocessing procedures of practical application, can be the independent utilization of a certain method in the said method, also can be that the combination of above-mentioned several method is used.By the cross correlation analytical approach, obtain correlation factors again.Correlation factors adopts the mutual correlation dimension computing method in the Chaotic Time Series Analysis, and computing formula is as follows:
Wherein,
Be correlation factors;
,
Be respectively case sample and analyzed sample spectrum data set.
For the unknown solid ferment process product to be measured sample, the tunning that equally at every turn takes by weighing about 40g is put into sample cup (spectrometer standard fitting) 1, sample cup 1 is positioned on the objective table 2, then the light that sends of the Halogen lamp LED in the near infrared spectrometer 6 shines on the sweat product sample through y-type optical fiber 3, and in the inner formation of this sample diffuse reflection, diffuse reflection light out enters near infrared spectrometer 6 through y-type optical fiber 3 again, and the spectral signal that obtains imports in the computing machine 4 by data line 5 after spectrometer 6 is analyzed conversion.In computing machine 4, finish the original spectrum pretreatment and with case sample spectral data correlation analysis, and the soft-sensing model that the correlation factors substitution that obtains has been established, the property value of corresponding key parameter index that just can the fast prediction sample to be tested, and be presented on the interface of computing machine 4.So far key parameter index attribute value that should the unknown sweat product to be measured sample is measured and is finished.
Claims (6)
1. based on the flexible measurement method of correlation analysis and ELM neural network, it is characterized in that: take different batches solid ferment process sample near infrared spectrum data correlation factors as the soft-sensing model input variable, take the actual measurement reference value of solid ferment process product parameter index as output variable, adopt the ELM neural network to carrying out solid ferment process key parameter soft sensor modeling; The step of described method is:
1) utilize diffuse reflection type near infrared spectra collection device to obtain the near infrared spectrum data of solid ferment process product, the spectral signal that is gathered imports computing machine into by data line after the spectrometer analysis conversion;
2) the original spectrum data that obtain are carried out pre-service, above-mentioned experiment is repeated N time, namely obtain the N batch data, choosing a collection of sweat monitoring near infrared spectrum data is the case sample data;
3) then adopt statistics correlation analysis and other (N-1) batch data to carry out correlation analysis;
4) utilize correlativity index analysis result and the actual measurement reference value of solid ferment process product parameter index to carry out related, set up the soft-sensing model based on ELM.
2. the flexible measurement method based on correlation analysis and ELM neural network according to claim 1, it is characterized in that: described ir data correlation factors analytical approach is chaos time sequence mutual correlation dimensional analysis method.
3. the flexible measurement method based on correlation analysis and ELM neural network according to claim 1 and 2, it is characterized in that: the concrete gatherer process of described step 1) is: collect N batch fermentation batch, different fermentations solid ferment process product sample constantly is used for carrying out model tuning, each sample takes by weighing puts into sample cup about 40g, and places it on the objective table; Near infrared spectrometer is connected with objective table by y-type optical fiber, and the spectral signal of collection imports near infrared spectrometer into by y-type optical fiber, is reached in the computing machine by the data line that is connected between computing machine and the spectrometer again.
4. the flexible measurement method based on correlation analysis and ELM neural network according to claim 1 and 2, it is characterized in that: the preprocess method described step 2) comprises standard normal variable conversion, level and smooth, centralization, differentiate, normalization and Wavelet Denoising Method, described preprocess method can be the independent utilization of a certain method in the described preprocess method, also can be that the combination of several method is used.
5. the flexible measurement method based on correlation analysis and ELM neural network according to claim 1 and 2, it is characterized in that: the actual measurement reference value in the described step 4) is measured by the conventional physical and chemical analysis method.
6. the flexible measurement method based on correlation analysis and ELM neural network according to claim 1 and 2, it is characterized in that: with reference to the concerned countries standard, record the reference measurement values of solid ferment process product parameter index by the physico-chemical analysis method, form a database, described parameter index comprises biomass content and/or protein content and/or humidity and/or PH.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012105416673A CN103018181A (en) | 2012-12-14 | 2012-12-14 | Soft measurement method based on correlation analysis and ELM neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012105416673A CN103018181A (en) | 2012-12-14 | 2012-12-14 | Soft measurement method based on correlation analysis and ELM neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103018181A true CN103018181A (en) | 2013-04-03 |
Family
ID=47967034
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012105416673A Pending CN103018181A (en) | 2012-12-14 | 2012-12-14 | Soft measurement method based on correlation analysis and ELM neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103018181A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103472197A (en) * | 2013-09-10 | 2013-12-25 | 江苏大学 | Cross-perception information interaction sensing fusion method in intelligent bionic evaluation for food |
| CN103593550A (en) * | 2013-08-12 | 2014-02-19 | 东北大学 | Pierced billet quality modeling and prediction method based on integrated mean value staged RPLS-OS-ELM |
| CN104330089A (en) * | 2014-11-17 | 2015-02-04 | 东北大学 | Map matching method by use of historical GPS data |
| CN105651727A (en) * | 2015-12-28 | 2016-06-08 | 中国计量学院 | Method for discriminating shelf life of apple through near infrared spectroscopy based on JADE and ELM |
| CN106290240A (en) * | 2016-08-29 | 2017-01-04 | 江苏大学 | A kind of method based on near-infrared spectral analysis technology to Yeast Growth curve determination |
| CN110873699A (en) * | 2018-08-30 | 2020-03-10 | 广东生益科技股份有限公司 | Method, device and system for online quality control of bonding sheet and storage medium |
| CN111072412A (en) * | 2019-12-27 | 2020-04-28 | 农业农村部规划设计研究院 | Numerical simulation test system for aerobic fermentation process |
| CN113298265A (en) * | 2021-05-22 | 2021-08-24 | 西北工业大学 | Heterogeneous sensor potential correlation learning method based on deep learning |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0793289A (en) * | 1993-06-18 | 1995-04-07 | Gold Star Co Ltd | Chaos processor |
| US6192273B1 (en) * | 1997-12-02 | 2001-02-20 | The Cleveland Clinic Foundation | Non-programmable automated heart rhythm classifier |
| CN1966934A (en) * | 2005-11-16 | 2007-05-23 | 中国石油大学(北京) | Method for prediction of collapse pressure and fracture pressure of stratum under drill bit while drilling |
| CN101140223A (en) * | 2007-08-29 | 2008-03-12 | 国际竹藤网络中心 | Textile fibre identification method |
| CN101339186A (en) * | 2008-08-07 | 2009-01-07 | 中国科学院过程工程研究所 | Method for on-line detection for solid-state biomass bioconversion procedure |
| CN101576467A (en) * | 2009-06-11 | 2009-11-11 | 哈尔滨工业大学 | In-situ determination method of fractal growth process of flocs in water |
| CN101630376A (en) * | 2009-08-12 | 2010-01-20 | 江苏大学 | Soft-sensing modeling method and soft meter of multi-model neural network in biological fermentation process |
| WO2011092549A1 (en) * | 2010-01-27 | 2011-08-04 | Nokia Corporation | Method and apparatus for assigning a feature class value |
| CN102521831A (en) * | 2011-12-02 | 2012-06-27 | 南京信息工程大学 | Robot vision image segmentation method based on multi-scale fractal dimension and neural network |
| CN102539375A (en) * | 2012-01-10 | 2012-07-04 | 江苏大学 | Straw solid-state fermentation process parameter soft measurement method and device based on near infrared spectrum |
| CN102737288A (en) * | 2012-06-20 | 2012-10-17 | 浙江大学 | Radial basis function (RBF) neural network parameter self-optimizing-based multi-step prediction method for water quality |
-
2012
- 2012-12-14 CN CN2012105416673A patent/CN103018181A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0793289A (en) * | 1993-06-18 | 1995-04-07 | Gold Star Co Ltd | Chaos processor |
| US6192273B1 (en) * | 1997-12-02 | 2001-02-20 | The Cleveland Clinic Foundation | Non-programmable automated heart rhythm classifier |
| CN1966934A (en) * | 2005-11-16 | 2007-05-23 | 中国石油大学(北京) | Method for prediction of collapse pressure and fracture pressure of stratum under drill bit while drilling |
| CN101140223A (en) * | 2007-08-29 | 2008-03-12 | 国际竹藤网络中心 | Textile fibre identification method |
| CN101339186A (en) * | 2008-08-07 | 2009-01-07 | 中国科学院过程工程研究所 | Method for on-line detection for solid-state biomass bioconversion procedure |
| CN101576467A (en) * | 2009-06-11 | 2009-11-11 | 哈尔滨工业大学 | In-situ determination method of fractal growth process of flocs in water |
| CN101630376A (en) * | 2009-08-12 | 2010-01-20 | 江苏大学 | Soft-sensing modeling method and soft meter of multi-model neural network in biological fermentation process |
| WO2011092549A1 (en) * | 2010-01-27 | 2011-08-04 | Nokia Corporation | Method and apparatus for assigning a feature class value |
| CN102521831A (en) * | 2011-12-02 | 2012-06-27 | 南京信息工程大学 | Robot vision image segmentation method based on multi-scale fractal dimension and neural network |
| CN102539375A (en) * | 2012-01-10 | 2012-07-04 | 江苏大学 | Straw solid-state fermentation process parameter soft measurement method and device based on near infrared spectrum |
| CN102737288A (en) * | 2012-06-20 | 2012-10-17 | 浙江大学 | Radial basis function (RBF) neural network parameter self-optimizing-based multi-step prediction method for water quality |
Non-Patent Citations (6)
| Title |
|---|
| LI HONGQIANG等: "Near-infrared spectroscopy with a fiber-optic probe for state variables determination in solid-state fermentation", 《PROCESS BIOCHEMISTRY》 * |
| 刘国海等: "近红外光谱结合ＥLＭ快速检测固态发酵过程参数pH值", 《光谱学与光谱分析》 * |
| 刘青格，陈斌: "基于相关分析技术的近红外光谱信息特征提取", 《农业机械学报》 * |
| 刘青格等: "相关检测技术在近红外光谱分析中的应用", 《光谱学与光谱分析》 * |
| 吴浩江等: "改进BP 神经网络在流型智能识别中的应用", 《西安交通大学学报》 * |
| 李春贵等: "一种识别混沌时间序列动力学异同性的方法", 《物理学报》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103593550A (en) * | 2013-08-12 | 2014-02-19 | 东北大学 | Pierced billet quality modeling and prediction method based on integrated mean value staged RPLS-OS-ELM |
| CN103472197A (en) * | 2013-09-10 | 2013-12-25 | 江苏大学 | Cross-perception information interaction sensing fusion method in intelligent bionic evaluation for food |
| CN103472197B (en) * | 2013-09-10 | 2015-03-04 | 江苏大学 | Cross-perception information interaction sensing fusion method in intelligent bionic evaluation for food |
| CN104330089A (en) * | 2014-11-17 | 2015-02-04 | 东北大学 | Map matching method by use of historical GPS data |
| CN104330089B (en) * | 2014-11-17 | 2017-12-29 | 东北大学 | A kind of method that map match is carried out using history gps data |
| CN105651727A (en) * | 2015-12-28 | 2016-06-08 | 中国计量学院 | Method for discriminating shelf life of apple through near infrared spectroscopy based on JADE and ELM |
| CN105651727B (en) * | 2015-12-28 | 2018-06-12 | 中国计量学院 | The method that near-infrared spectrum analysis based on JADE and ELM differentiates apple shelf life |
| CN106290240A (en) * | 2016-08-29 | 2017-01-04 | 江苏大学 | A kind of method based on near-infrared spectral analysis technology to Yeast Growth curve determination |
| CN110873699A (en) * | 2018-08-30 | 2020-03-10 | 广东生益科技股份有限公司 | Method, device and system for online quality control of bonding sheet and storage medium |
| CN111072412A (en) * | 2019-12-27 | 2020-04-28 | 农业农村部规划设计研究院 | Numerical simulation test system for aerobic fermentation process |
| CN113298265A (en) * | 2021-05-22 | 2021-08-24 | 西北工业大学 | Heterogeneous sensor potential correlation learning method based on deep learning |
| CN113298265B (en) * | 2021-05-22 | 2024-01-09 | 西北工业大学 | Heterogeneous sensor potential correlation learning method based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103018181A (en) | Soft measurement method based on correlation analysis and ELM neural network | |
| CN102539375A (en) | Straw solid-state fermentation process parameter soft measurement method and device based on near infrared spectrum | |
| CN101210875A (en) | Damage-free measurement method for soil nutrient content based on near infrared spectra technology | |
| CN104931470A (en) | Fluorescence hyperspectral technology-based pesticide residue detection device and method | |
| CN102778442B (en) | Method for rapidly identifying type of balsam material liquid for cigarette | |
| CN101915738A (en) | Method and device for rapidly detecting nutritional information of tea tree based on hyperspectral imaging technique | |
| CN103134765A (en) | Chinese medicine sample authenticity preliminary screening method based on terahertz time-domain spectrum | |
| CN103278503B (en) | Multi-sensor technology-based grape water stress diagnosis method and system therefor | |
| CN103353446A (en) | Method of near-infrared rapid detection of physicochemical indexes in wine | |
| CN102876816A (en) | Fermentation process statue monitoring and controlling method based on multi-sensor information fusion | |
| CN103411906A (en) | Near infrared spectrum qualitative identification method of pearl powder and shell powder | |
| CN105092579A (en) | Mango quality non-destructive testing device | |
| CN104034691A (en) | Rapid detection method for beta vulgaris quality | |
| CN106290240A (en) | A kind of method based on near-infrared spectral analysis technology to Yeast Growth curve determination | |
| CN201503392U (en) | Handheld soil nutrient nondestructive measurement device based on near infrared spectrum | |
| CN111537469A (en) | Apple quality rapid nondestructive testing method based on near-infrared technology | |
| CN103743705A (en) | Rapid detection method for sorghum halepense and similar species | |
| CN103234923A (en) | On-line monitoring method of total sugar content during yellow wine fermentation process | |
| CN111562235A (en) | Method for rapidly identifying black-leaf outbreak disease and infection degree of tobacco leaves based on near infrared spectrum | |
| CN111795943A (en) | Method for nondestructive detection of exogenous doped sucrose in tea based on near infrared spectrum technology | |
| CN102313715A (en) | Method for detecting honey quality base on laser technology | |
| CN103267740A (en) | Straw fermentation process characteristic wave number soft instrument apparatus and construction method thereof | |
| CN103487398A (en) | Analysis method of lysine fermentation liquid | |
| CN103048275A (en) | Adaptive soft-instrument device and adaptive soft-instrument construction method based on near infrared spectrum | |
| CN210775225U (en) | Fruit maturity detection and picking device based on Raman spectrum |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20130403 |